KR100850810B1 - Method and apparatus for variable accuracy inter-picture timing specification for digital video encoding - Google Patents

Abstract

A method and apparatus for variable accuracy inter-picture timing specification for digital video encoding is disclosed. Specifically the present invention discloses a system that allows the relative timing of nearby video pictures to be encoded in a very efficient manner. In one embodiment, the display time difference between a current video picture (105) and a nearby video picture is determined. The display time difference is then encoded (180) into a digital representation of the video picture. In a preferred embodiment, the nearby video picture is the most recently transmitted stored picture. For coding efficiency, the display time difference may be encoded using a variable length coding system or arithmetic coding. In an alternate embodiment, the display time difference is encoded as a power of two to reduce the number of bits transmitted.

Description

METHOD AND APPARATUS FOR VARIABLE ACCURACY INTER-PICTURE TIMING SPECIFICATION FOR DIGITAL VIDEO ENCODING}             

The present invention relates to the field of multimedia compression systems. In particular, the present invention discloses a method and system for specifying variable accuracy inter-picture timing.

The digital-based electronic media format is finally about to replace the analog electronic media format. Digital compact discs (CD) have long replaced analog vinyl records. Analog magnetic cassette tapes are increasingly being used. Now, second and third generation digital audio systems, such as minidiscs and MPEG audio-layer 3 (MP3), are occupying the market for compact discs in first generation digital audio formats.

Video media has been slowly changed to digital storage and transmission formats rather than audio media. This was mainly due to the huge amount of digital information needed to accurately represent video in digital form. The enormous amount of digital information needed to accurately represent video requires high capacity digital storage systems and wide bandwidth transmission systems.

However, video is now rapidly changing to digital storage and transmission formats. High-speed computer processors, high-density storage systems, and efficient new compression and encoding algorithms have finally made digital video practical at the price of consumers. Digital Versatile Discs (DVDs), or digital video systems, are among the fastest selling consumer electronics products of many years. DVD has been rapidly replacing videocassette recorders (VCRs) as a playback system for pre-recorded video, selected because of its high video quality, very high audio quality, convenience, and other features. Old analog NTSC (National Television Standards Committee) video transmission systems are currently being replaced by digital Advanced Television Standards Committee (ATSC) video transmission systems.

Computer systems have been using a variety of different digital video encoding formats for many years. The best digital video compression and encoding system used by computer systems has been digital video systems supported by a group of motion picture experts, commonly known as acronym MPEG. The three best known and widely used MPEG digital video formats are simply known as MPEG-1, MPEG-2, and MPEG-4. Video CDs and earlier consumer-grade digital video editing systems use the earlier MPEG-1 digital video encoding format. Digital Versatile Disc (DVD) and Dish Network brand DBS (Direct Broadcast Satellite) television broadcasting systems use a higher quality MPEG-2 digital video compression and encoding system. MPEG-4 encoding systems are rapidly being adapted by modern computer-based digital video encoders and related digital video players.

The MPEG-2 and MPEG-4 standards compress a series of video frames or video fields and then encode the compressed frames or fields into a digital bitstream. When encoding a video frame or field with the MPEG-2 and MPEG-4 systems, the video frame or field is divided into a rectangular grid of macro blocks. Each macro block is independently compressed and coded.

When compressing video frames or fields, the MPEG-4 standard uses three types of frames or fields: intra-frames (I-frames), unidirectional predictive frames (P-frames), or bidirectional prediction. It can be compressed into one of the frames (B-frames). Intra-frames encode independent video frames completely independently without reference to other video frames. P-frames define a video frame with reference to a single previously displayed video frame. A B-frame defines a video frame by referring to a video frame displayed before the current frame and a video frame to be displayed after the current frame. Due to the efficient use of redundant video information, P-frames and B-frames generally provide the best compression.

Methods and apparatus for specifying variable accuracy inter-picture timing for digital video encoding are disclosed. Specifically, the present invention discloses a system that allows the relative timing of adjacent video pictures to be encoded very efficiently. In one embodiment, the display time difference between the current video picture and the adjacent video picture is determined. The display time difference is then encoded into a digital representation of the video picture. In a preferred embodiment, the adjacent video picture is the most recently transmitted, stored picture.

For coding efficiency, the display time difference may be encoded using a variable length coding system or an arithmetic coding system. In an alternative embodiment, the display time difference is encoded as the power of two to reduce the number of bits transmitted.

Other objects, features, and advantages of the present invention will become apparent from the accompanying drawings and the following detailed description.

The objects, features and advantages of the present invention will become apparent to those skilled in the art from the following detailed description.

1 shows a high level block diagram of one possible digital video encoder system.

2 is a diagram showing a series of video pictures in the order in which they should be displayed, with arrows connecting different pictures indicating inter-picture dependencies created using motion compensation.

3 is a view of the video picture of FIG. 2 listed in the preferred picture transfer order, with arrows connecting different pictures to indicate inter-picture dependencies created using motion compensation.

4 is a graphical representation of a series of video images, selected such that the distance between the video images that reference each other is a power of two.

A method and system for specifying variable accuracy inter-picture timing in a multimedia compression and coding system is disclosed. In the following description, special names are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to one skilled in the art that this particular description is not necessary to practice the invention. For example, the present invention has been described with reference to an MPEG-4 multimedia compression and encoding system. However, the same technique can be easily applied to other types of compression and encoding systems.

Multimedia Compression and Encoding Overview

1 is a high level block diagram of a general digital video encoder 100 well known in the art. The digital video encoder 100 receives an input video stream 105 of video frames on the left side of the block diagram. Each video frame is processed by a discrete cosine transform (DCT) unit 110. Frames may be processed independently (intra-frames) or with reference to information from other frames received from the motion compensation unit (inter-frames). Next, the quantizer (Q) unit 120 quantizes the information from the discrete cosine transform unit 110. Finally, the quantized video frame is encoded by entropy encoder (H) unit 180 to generate an encoded bitstream. The entropy encoder (H) unit 180 may use a variable length coding (VLC) system.

Since inter-frame coded video frames are defined with reference to other adjacent video frames, the digital video encoder 100 needs to generate a copy indicating how each frame appears decoded within the digital video decoder, Thereby, the frame can be encoded. Thus, the lower part of the digital video encoder 100 is actually a digital video decoder system. Specifically, inverse quantizer (Q- ¹ ) unit 130 performs inverse of quantization of video frame information, and inverse discrete cosine transform (DCT- ¹ ) unit 140 performs inverse of discrete cosine transform of video frame information. Do it. After all the DCT coefficients are reconstructed from the iDCT, the motion compensation unit uses that information along with the motion vector to reconstruct the encoded frame used as a reference frame for motion estimation of the next frame.

The decoded video frame can then be used to encode an inter-frame (P-frame or B-frame) defined in relation to the information of the decoded video frame. Specifically, motion compensation (MC) unit 150 and motion estimation (ME) unit 160 are used to determine motion vectors and to generate differential values used to encode inter-frames.                 

A rate controller 190 receives information from many other components in the digital video encoder 100 and uses that information to allocate a bit budget of each video frame. The rate controller 190 must allocate the bit budget in such a way as to produce the highest quality digital video bit stream in accordance with the specified set of restrictions. Specifically, the rate controller 190 can either display an overflow buffer (which exceeds the amount of available memory in the decoder by sending more information than can be stored) or a video frame to display due to not transmitting the video frame fast enough. We want to generate the highest quality compressed video stream without an underflow buffer.

Multimedia Compression and Encoding Overview

In some video signals, the time between successive video pictures (frames or fields) may not be constant (note: the term video picture is used herein to generically refer to a video frame or video field). For example, some video pictures may be broken due to transmission bandwidth constraints. Video timing may also vary due to camera irregularities or special effects such as slow motion or high speed motion. In some video streams, the original video source may simply have non-uniform inter-picture time by design. For example, since no video timing is generated by a uniform video capture system such as a video camera system, composite video such as computer graphics animation may have non-uniform timing. Flexible digital video coding systems must be able to handle uneven timing.

Many digital video encoding systems divide a video picture into a rectangular grid of macro blocks. Each individual macro block from the video picture is compressed and encoded independently. In some embodiments, subblocks of macro blocks known as 'pixel blocks' are used. Such a pixel block may have its own motion vector that can be interpolated. The invention is equally applicable to both macroblocks and pixelblocks, but reference is made herein to macroblocks.

Some video coding standards, such as the ISO MPEG standard or the ITU H.264 standard, use different types of predicted macro blocks to encode video pictures. In one scenario, the macro block can be one of three types:

1. I-Macro Block-An intra (I) macro block does not use information from any other video picture in its encoding (this I macro block is fully self-defined).

2. P-Macro Block-One-way prediction (P) macro block refers to picture information from one previous video picture.

3. B-Macro Block-Bidirectional Prediction (B) Macro block uses information from one previous picture and one future video picture.

If all macro blocks in a video picture are intra-macro blocks, the video picture is an intra-frame. If the video picture contains only unidirectional predicted macro blocks or intra-macroble, the video picture is known to be a P-frame. If the video picture contains some bi-predicted macro block, the video picture is known to be a B-frame. For simplicity, this specification considers the case where all macro blocks in a given picture are of the same type.

An exemplary sequence of video pictures to be encoded is

I ₁ B ₂ B ₃ B ₄ P ₅ B ₆ B ₇ B ₈ B ₉ P ₁₀ B ₁ l P ₁₂ B ₁₃ I ₁₄ ...

Where the letter (I, P, or B) indicates whether the video picture is an I-frame, or a P-frame, or a B-frame, and the number indicates the camera order of the video picture in the sequence of video pictures. Indicates. The camera order is the order in which the camera recorded the video pictures, and therefore also the order in which the video pictures should be displayed (display order).

The previous exemplary series of video pictures is shown graphically in FIG. 2. Referring to Fig. 2, the arrow indicates that macro blocks from stored pictures (in this case I-frames or P-frames) are used for motion compensation prediction of other pictures.

In the scenario of FIG. 2, information from other pictures is not used for encoding the intra-frame video picture I ₁ . Video picture P ₅ is a P-frame that uses the video information from the previous video picture I ₁ in its encoding, so an arrow can be drawn from video picture I ₁ to video picture P ₅ . Video picture B ₂ , video picture B ₃ , and video picture B ₄ all use the information from video picture I ₁ and video picture P ₅ in their encoding, so that an arrow from video picture I ₁ and video picture P ₅ . Can be drawn as a video picture B ₂ , a video picture B ₃ , and a video picture B ₄ . As mentioned above, the time between images is generally not the same.

Since the B-picture uses information from future pictures (pictures to be displayed later), the transmission order (i) is usually different from the display order. Specifically, the video picture necessary to compose another video picture must be transmitted first. In the case of the sequence, the transmission order is

I ₁ P ₅ B ₂ B ₃ B ₄ P ₁₀ B ₆ B ₇ B ₈ B ₉ P ₁₂ B ₁₁ I ₁₄ B ₁₃ ...

Can be.

3 graphically illustrates the transmission sequence of the video image of FIG. 2. Further, the arrows in the figure indicate that macro blocks from stored video pictures (in this case I or P) are used for motion compensation prediction of other video pictures.

Referring to FIG. 3, the system first transmits an I-frame I ₁ that does not depend on another frame. The system then transmits a P-frame video picture P ₅ that depends on the video picture I ₁ . Next, the system even if the video picture B ₂ to be displayed before video picture P ₅ and transmits the B- frame video picture B ₂ after the video picture P _5. This is because, when it is time to encode B ₂ , the decoder has already received and stored the information of the video pictures I ₁ and P ₅ required to decode the video picture B ₂ . Similarly, video pictures I ₁ and P ₅ are prepared for use in decoding subsequent video pictures B ₃ and video pictures B ₄ . The receiver / decoder rearranges the video picture sequence for proper display. In this operation, I pictures and P pictures are often referred to as stored pictures.

Coding of P-frame pictures generally uses motion compensation, and the motion vector of each macro block of the picture is calculated. Using the calculated motion vector, a predictive macro block (P macro block) may be formed by the transformation of the pixels of the previous picture as described above. Next, the difference between the actual macroblock and the predictive macroblock of the P-frame picture is encoded for transmission.

Each motion vector may also be transmitted via predictive coding.

For example, motion vector prediction may be formed using adjacent motion vectors. In such a case, the difference between the actual motion vector and the motion vector is encoded for transmission.

Each B-macro block uses two motion vectors, a first motion vector referring to the previous video picture described above and a second motion vector referring to a future video picture. From these two motion vectors, two prediction macro blocks are calculated. Next, the two prediction macro blocks are combined together using some function to form the final prediction macro block. As above, the difference between the actual macroblock and the last predicted macroblock in the B-frame picture is encoded for transmission.                 

As in the case of the P-macro block, each motion vector (MV) of the B-macro block may be transmitted through predictive coding. Specifically, the predicted motion vector is formed using the adjacent motion vector. Next, the difference between the actual motion vector and the predicted motion vector is encoded for transmission.

However, in the case of a B-macro block, there is an opportunity to interpolate the motion vector from the motion vector in the closest stored picture macro block. Such interpolation is performed in digital video encoders and digital video decoders.

This motion vector interpolation works particularly well for video pictures from video sequences when the camera pans at low speed across a fixed background. In practice, such motion vector interpolation may be good enough to be used alone. Specifically, this means that it is not necessary to calculate or transmit the difference information of these B-macro block motion vectors encoded using interpolation.

For further explanation, in the above scenario, denote an inter-picture display time between pictures i and j as D _{i, j} . That is, the display times of the image are each If T _i and T _j ,

D _{i, j} = T _i -T _j

From this expression,

D _{i, k} = D _{i, j} + D _{j, k}

D _{i, k} = -D _{k, i}

to be. Note that D _{i, j} may be negative in some cases.

Thus, if MV _{5, l} is a motion vector of the P ₅ macroblock referenced to I _l , then for a corresponding macroblock of B ₂ , B ₃ and B ₄ , the motion vector referenced to I ₁ and P ₅ is

MV _2,1 = MV _5,1 * D _2,1 / D _5,1

MV _5,2 = MV _5,1 * D _5,2 / D _5,1

MV _3,1 = MV _5,1 * D _3,1 / D _5,1

MV _5,3 = MV _5,1 * D _5,3 / D _5,1

MV _4,1 = MV _5,1 * D _4,1 / D _5,1

MV _5,4 = MV _5,1 * D _5,4 / D _5,1

Are interpolated by. Note that since the ratio of display times is used for motion vector prediction, no absolute display time is needed. Thus, the relative display time can be used as the D _{i, j} display time value.

This scenario can be generalized as, for example, in the H.264 standard. In generalization, a P or B picture may use a previously transmitted picture for motion vector prediction. Thus, in the above case, picture B ₃ can use picture I ₁ and picture B ₂ for its prediction. Also, the motion vector can be extrapolated without just interpolating. So in this case,

MV _3.1 = MV _2,1 * D _3,1 / D _2,1

Such motion vector extrapolation (or interpolation) may also be used in the prediction process for predictive encoding of motion vectors.

In any case, the problem in the case where the inter-picture time is uneven is to transmit the relative display time value of D _{i, j} to the receiver, which is the subject of the present invention. In one embodiment of the invention, for each picture after the first picture, the display time difference between the current picture and the most recently transmitted stored picture is transmitted. For error resilience, the transmission can be repeated many times in the picture, for example in the slice header of the so-called MPEG or H.264 standard. If all slice headers are lost, other pictures that depend on the missing picture, perhaps as decoding information, cannot be decoded.

Thus, in this scenario,

D _5,1 D _2,5 D _3,5 D _4,5 D _10,5 D _6,10 D _7,10 D _8,10 D _9,10 D _12,10 D _11,12 D _14,12 D _{13 , 14} ...

Send it.

For motion vector estimation, the accuracy requirement for D _{i, j} may vary from picture to picture. For example, if there is only a single B-frame picture B ₆ between two P-frame pictures P ₅ and P ₇ ,

D _7,5 = 2 and D _6,7 = -1

It is enough to send only. Here, the D _{i, j} display time value is a relative time value. Instead, if video picture B ₆ is only one quarter of the distance between video picture P ₅ and video picture P ₇ , then the appropriate D _{i, j} display time value to be transferred is

D _7,5 = 4 and D _6,7 = -1

to be. In the two previous examples, the display time between video picture B ₆ and video picture video picture P ₇ is used as the display time “unit”, and the display time difference between video picture P ₅ and picture video picture P ₇ is four display time “units”. "Be careful.

In general, motion vector estimation is less complex if the divisor is a power of two. This is easily accomplished in this embodiment if D _{i, j} (inter-picture time) between two stored pictures is selected as a power of 2 as shown graphically in FIG. Alternatively, the estimation procedure may be defined to truncate or round all divisors to be powers of two.

In the case where the inter-picture time is a power of two, the number of data bits can be reduced if only (two) integer squares are transmitted instead of the full value of the inter-picture time. 4 graphically illustrates the case where the distance between images is selected to be a power of two. In such a case, the D _3,1 display time value of 2 between the video picture P ₁ and the picture video picture P ₃ is transmitted as ¹ (since 2 ¹ = 2), and between the picture picture P ₇ and the picture video picture P ₃ A D _7,3 display time value of 4 may be sent as 2 (since 2 ² = 4).

In some cases, motion vector interpolation may not be used. However, it is still necessary to transmit the display order of the video pictures to the receiver / player system so that the receiver / player system can display the video pictures in the proper order. In this case, D _{i, j} Simple signed integer values are sufficient regardless of the actual display time. In some applications, only a sign may be needed.

The inter picture time D _{i, j} can simply be transmitted as a simple signed integer value. However, many methods can be used to encode the _{Di, j} values to achieve additional compression. For example, code bits followed by variable length code sizes are relatively easy to implement and provide coding efficiency.

One such variable length coding system that can be used is known as Universal Variable Length Code (UVLC). The UVLC variable length coding system has a code word:

1 = 1

2 = 0 1 0

3 = O 1 l

4 = 0 0 1 0 0                 

5 = 0 0 1 0 1

6 = 0 0 1 1 0

7 = 0 0 1 1 1

8 = 0 0 0 1 0 0 0 ...

Is given by

Another method of encoding the inter-picture time may be to use arithmetic coding. In general, arithmetic coding uses conditional probabilitie to achieve very high compression of data bits.

As such, the present invention introduces a simple but powerful method of encoding and transmitting the inter-picture display time. The encoding of the inter-picture encoding display time can be made very efficient by using variable length coding or arithmetic coding. Also, the desired accuracy can be selected to meet the needs of the video decoder, but no more is needed.

In the above, the system for specifying the variable accuracy inter-picture timing in the multimedia compression and coding system has been described. It is contemplated that those skilled in the art will be able to modify and modify the materials and arrangements of the components of the invention without departing from the scope of the invention.

Claims (93)

delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete A method relating to a stream comprising a first video picture, a second video picture and a third video picture, (i) a first order difference between the order value relating to the third video picture and the order value relating to the first video picture, and (ii) the order value relating to the second video picture and the first video. Obtaining a specific value based on a second order difference between the order values relating to the image, Obtaining a specific motion vector relating to the second video picture based on the motion vector relating to the third video picture and the specific value How to include. 68. The method of claim 67, wherein the order value for the particular video picture is for specifying the position of the particular video picture within a sequence of video pictures. 68. The method of claim 67, wherein the order value for the particular video picture is indicative of the positional relationship of the particular video picture with respect to another video picture. 70. The method of claim 69, wherein the another video picture is an I video picture. 68. The method of claim 67, wherein the specified value is proportional to the second order difference value. 68. The method of claim 67, wherein the specified value is inversely proportional to the first order difference value. 68. The method of claim 67, wherein the order value for the second video picture is obtained from a value stored in a slice header associated with the second video picture. 68. The method of claim 67, further comprising obtaining the first order difference value and obtaining the second order difference value before obtaining the specific value. 68. The method of claim 67, further comprising decoding at least one video picture using the obtained motion vector. 68. The method of claim 67, wherein the order value of the particular video picture represents a time value. A computer-readable storage medium for processing a first video picture, a second video picture, and a third video picture and storing a computer program executable by at least one processor, the computer program comprising: (i) a first order difference value between the order value for the third video picture and the order value for the first video picture, and (ii) the order value for the second video picture and the first video picture. Obtaining a specific value based on a second order difference value between the order values, Obtaining a specific motion vector relating to the second video picture based on the motion vector relating to the third video picture and the specific value And a set of instructions relating to the computer readable storage medium. 78. The computer readable storage medium of claim 77, wherein the order value for a particular video picture is for specifying a location of the particular video picture within a sequence of video pictures. 78. The computer readable storage medium of claim 77, wherein the order value for a particular video picture is indicative of a positional relationship of the particular video picture with respect to another video picture. 80. The computer readable storage medium of claim 79, wherein the another video picture is an I video picture. 78. The computer readable storage medium of claim 77, wherein said particular value is proportional to said second order difference value. 78. The computer readable storage medium of claim 77, wherein said specified value is inversely proportional to said first order difference value. 78. The computer readable storage medium of claim 77, wherein the order value for the second video picture is obtained from a value stored in a slice header associated with the second video picture. 78. The computer program of claim 77, wherein the computer program further comprises a set of instructions for obtaining the first order difference value and obtaining the second order difference value prior to obtaining the particular value. Storage media. 78. The computer program product of claim 77, wherein the computer program further comprises a set of instructions for decoding at least one video picture using the obtained motion vector. A computer-readable storage medium for processing a first video picture, a second video picture, and a third video picture and storing a computer program executable by at least one processor, the computer program comprising: (i) a first order difference value indicating a difference between the order value relating to the third video picture and the order value relating to the first video picture, and (ii) the order value relating to the second video picture and the first video. Obtaining a specific value based on a second order difference value indicating a difference between the order values with respect to the picture, wherein the order value with respect to the particular video picture indicates the position of the particular video picture within a sequence of video pictures; Obtaining a motion vector relating to the second video picture based on the motion vector relating to the third video picture and the specific value; Obtaining another motion vector relating to the second video picture based on the motion vector relating to the third video picture And a set of instructions relating to the computer readable storage medium. 87. The computer readable storage medium of claim 86, wherein the order value for the particular video picture specifies the display order of the particular video picture within a sequence of video pictures. 87. The computer readable storage medium of claim 86, wherein the computer program further comprises a set of instructions for decoding at least one video picture using one of the obtained motion vectors. A method for a stream comprising a first video picture, a second video picture, and a third video picture, the method comprising: (i) an order value relating to the third video picture and an order relating to the first video picture A first order difference value representing a difference between the search values and (ii) the second video picture. A second order difference indicating a difference between an order value and an order value relating to the first video picture Obtaining a specific value based on the value, wherein an order value relating to the specific video picture Indicates the position of the particular video picture within the sequence of video pictures; Based on the motion vector and the specific value relating to the third video picture; Obtaining a motion vector relating to the second video picture, The second video based on the motion vector with respect to the third video picture Obtaining another motion vector with respect to the picture. 89. The method of claim 89, wherein the order value for the particular video picture specifies the display order of the particular video picture within a sequence of video pictures. 90. The method of claim 89, further comprising decoding at least one video picture using one of the obtained motion vectors. 78. The computer program of claim 77, wherein said computer program further comprises a set of instructions for calculating said first order difference value and calculating said second order difference value before obtaining said particular value. Storage media. 78. The computer readable storage medium of claim 77, wherein the order value for the particular video picture represents a time value.

Images